Estimating overdiagnosis in giant cell arteritis diagnostic pathways using genetic data: genetic association study

Abstract

Objectives: Prompt diagnosis of giant cell arteritis (GCA) is important to avert visual loss. False-negative temporal artery biopsy (TAB) can occur. Without vascular imaging, GCA may be overdiagnosed in TAB-negative cases, but it is unclear how often this occurs. An unbiased test is a way to address an imperfect reference standard. We used the known Human Leukocyte Antigen (HLA) region genetic association of TAB-positive GCA to estimate the extent of overdiagnosis before widespread adoption of temporal artery ultrasound as a first-line test. Methods: Patients diagnosed with GCA between 1990-2014 consented to the UKGCA Consortium study. HLA region variants were jointly imputed from genome-wide genotypic data of cases and controls. Per-allele frequencies across all variants with p<1.0x10-5 were compared with population control data to estimate overdiagnosis rates in cases without a positive TAB. Results: Genetic data from 663 patients diagnosed with GCA were compared with data from 2619 population controls. TAB-negative GCA (n=147) and GCA without a TAB result (n=160) had variant frequencies intermediate between those of TAB-positive GCA and population controls. Making several strong assumptions, we estimated that around two-thirds of TAB-negative cases and around one-third of cases without TAB result may have been overdiagnosed. From these data, TAB sensitivity is estimated at around 88%. Conclusions: Conservatively assuming 95% specificity, TAB has a negative likelihood ratio of around 0.12. Genotyping alone cannot diagnose GCA at the individual level. Group-level HLA variant genotyping might be used to compare the overall accuracy of different diagnostic pathways or different classification criteria sets.